Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875

Contents lists available at ScienceDirect Neuroscience and Biobehavioral Reviews

journal homepage: www.elsevier.com/locate/neubiorev

Review Rodent empathy and affective neuroscience

Jules B. Panksepp ∗, Garet P. Lahvis ∗∗

Department of Behavioral Neuroscience, Oregon Health and Science University, 3181 SW Sam Jackson Park Road, L470, Portland, OR 97239, USA article info abstract

Article history: In the past few years, several experimental studies have suggested that empathy occurs in the social Received 11 October 2010 lives of rodents. Thus, rodent behavioral models can now be developed to elucidate the mechanistic Received in revised form 24 May 2011 substrates of empathy at levels that have heretofore been unavailable. For example, the finding that Accepted 27 May 2011 mice from certain inbred strains express behavioral and physiological responses to conspecific distress, while others do not, underscores that the genetic underpinnings of empathy are specifiable and that Keywords: they could be harnessed to develop new therapies for human psychosocial impairments. However, the Fear advent of rodent models of empathy is met at the outset with a number of theoretical and semantic Pain Distress problems that are similar to those previously confronted by studies of empathy in humans. The distinct Emotion underlying components of empathy must be differentiated from one another and from lay usage of the Social behavior term. The primary goal of this paper is to review a set of seminal studies that are directly relevant to Mus musculus developing a concept of empathy in rodents. We first consider some of the psychological phenomena Rattus Norvegicus that have been associated with empathy, and within this context, we consider the component processes, Reciprocity or endophenotypes of rodent empathy. We then review a series of recent experimental studies that Altruism demonstrate the capability of rodents to detect and respond to the affective state of their social partners. We focus primarily on experiments that examine how rodents share affective experiences of fear, but we also highlight how similar types of experimental paradigms can be utilized to evaluate the possibility that rodents share positive affective experiences. Taken together, these studies were inspired by Jaak Panksepp’s theory that all mammals are capable of felt affective experiences. © 2011 Elsevier Ltd. All rights reserved.

Contents

1. Introduction ...... 1864 1.1. Empathy ...... 1865 1.2. Foundational studies ...... 1866 2. Social modulation of pain ...... 1867 3. Social interaction with a recently distressed conspecific modulates subsequent learning about environmental signals that predict distress. 1867 4. Social transfer of fear ...... 1869 5. Synopsis and conclusions ...... 1871 References ...... 1873

1. Introduction writings of primatologist Franz de Waal, more principled and evo- lutionary based perspectives on empathy have emerged (de Waal, Historically, empathy has been considered a high-level affec- 2008). For instance, it is now generally accepted that many primate tive/cognitive process that is expressed exclusively by humans. species have a capacity for empathy, that empathy can be a proxi- However, recent scientific developments have placed this anthro- mate mechanism underlying the expression of altruistic behavior, pocentric view into question. Prompted in part by the research and and that empathy is the product of an integrated set of brain pro- cesses. Moreover, contemporary views of empathy consider its expression to be a product of several behavioral, affective and cog- ∗ nitive processes, each of which can vary with development, context Corresponding author. Tel.: +1 503 494 8286; fax: +1 503 494 6877. ∗∗ and species (see below). Deconstructing empathy into specifiable Corresponding author. Tel.: +1 503 346 0820; fax: +1 503 494 6877. E-mail addresses: [email protected] (J.B. Panksepp), [email protected] components is useful elucidating the biological substrates that con- (G.P. Lahvis). tribute to impairments in social interaction.

0149-7634/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.neubiorev.2011.05.013 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1865

In this paper, we consider one core feature of empathy in rodents response, which allowed individuals to relate with the emotional – the ability to share affective experiences. We review and compare state of others even though they did not necessarily share the same the experimental approaches and results of a series of recent papers state (for an excellent review of different definitions of empathy that explored different aspects of rodent behavior in response to the and its historical evolution see Davis, 1994). distress of conspecifics. Rather than reviewing research on empathy Theoretical developments in empathy research now view the in primates (Davis, 1994; Farrow and Woodruff, 2007; Silk, 2007; expression of empathy as the result of an interaction between sev- Decety, 2010), we examine how shared affect can be modeled in eral component processes, both affective and cognitive (Hoffman, laboratory rats and mice. In this regard, studies of shared affect in 1987; Preston and de Waal, 2002; Decety and Jackson, 2004). rodents can provide a level of biological resolution that has never Preston and de Waal (2002) hypothesized that these processes uti- been achieved in empathy research. lize an ancient perception–action coupling mechanism in which a Several of the studies that we describe employ a fear- subject’s attention to the ‘state’ of another can automatically acti- conditioning paradigm in which an individual is challenged to learn vate the same state in the subject. Regardless of the underlying an association between a conditioned stimulus (CS), such as a tone mechanism, this model serves as a very useful heuristic insofar or context, and an unconditioned stimulus (UCS), such as a deliv- that viewing empathy as a psychological phenomenon which stems ery of a shock. In a standard fear-conditioning paradigm, a subject from several underlying processes offers a practical strategy for learns to associate the CS with a UCS that engenders pain. However, employing biological approaches in empathy research. In the rest many of the studies described herein utilize a very different UCS, of this section, we review what some of these component pro- a distress cue that has been generated by the induction of pain in cesses are in attempt to clarify our own hypothesis that rodents another individual (Section 4). Consistent with the affective neuro- are capable of sharing affective experiences. science approach (Panksepp, 1998), we adopt the perspective that Emotional contagion is a psychological process that is relevant such associations can be learned because changes in an animal’s to empathy and refers to a phenomenon in which the perception of subjective state occur while others are undergoing distress. In this a behavioral change in an individual appears to automatically acti- scenario, a subject’s subsequent responsiveness to a CS thus reflects vate the same process in another individual. Emotional contagion its previous emotional experiences with a conspecific in pain. is thus a reflexive behavioral process among individuals within the A basic premise of affective neuroscience is that careful behav- context of a motivationally salient event. By definition, contagion ioral and neuroanatomical manipulations in the laboratory can requires that two individuals contemporaneously express a behav- yield insights into how affective and cognitive processes are dis- ior that reflects a common experience. Perhaps the most common tinct both in terms of their overt expression and their respective examples of emotional contagion in humans are infectious crying neural circuitries (Panksepp, 2005). Through the lens of affective among babies and yawning among adults. Although emotional con- neuroscience, we can envision building a robust framework for tagion fits within a more generally accepted definition of empathy; elucidating the neurobiological substrates that underlie different “the generation of an affective state more appropriate to the sit- aspects of empathy in animals. uation of another compared to one’s own” (Hoffman, 1975), it is excluded from others, particularly when there is an emphasis on 1.1. Empathy the ability to distinguish self from other. Importantly, emotional contagion does not require an ability to discern whether the source The study of empathy is heavily influenced by questions about of an affective experience comes from one’s self or from another terminology, and there continues to be an imprecise and there- individual (Singer and Lamm, 2009). fore confusing usage of the term ‘empathy’ both in the scientific The ability to distinguish self from other is a key feature of emo- literature and among the lay public. In this paper, we will not tional empathy, in which directed attention to another’s emotional review this literature or how empathy is expressed in humans state can lead to the same state in a subject. Thus, like emotional (readers are referred to MacLean, 1967; Hoffman, 1981; Davis, contagion, emotional empathy involves ‘state-matching’ between 1994; Decety, 2010). Rather, we will point out a few salient def- individuals. However, emotional empathy is exclusively concerned initions that can be useful in developing a concept of empathy in with the affective state of individuals, as opposed to a reflex- rodents. ive behavioral response. Importantly, because emotional empathy The word empathy has undergone a substantial evolution in requires an ability to distinguish one’s self from another, it can sub- the last century. Lipps (1903) provided the original definition of sequently lead to helping behaviors and also can be engaged by empathy as a process by which “the perception of an emotional one’s personal recollection of an experience (Davis, 1994; Preston gesture in another directly activates the same emotion in the per- and de Waal, 2002). ceiver, without any intervening labeling, associative or cognitive While emotional empathy inherently requires that two indi- perspective-taking processes.” (Preston and de Waal, 2002, pp. 2). viduals share an affective experience, its expression also can During the next one-hundred years, perspectives on empathy were be modulated by cognitive processes. These cognitive aspects of substantially expanded and refined. Some investigators focused on empathy incorporate changes in the emotional state of one individ- how individuals perceive and respond to the emotional expressions ual that are subsequently experienced by another individual after of others. Such approaches emphasized a role for affective arousal, some degree of additional processing (e.g., top–down processing). associative learning and motor mimicry (i.e., imitation). More cog- Examples of this include contextual appraisals, such as integrating nitive approaches to studying empathy were based on describing past experiences or familiarity, and high-level cognitive phenom- how an individual comes to understand the perspective of another ena, such as perspective taking (see below). For instance, in some by actively projecting into the psychology of their social partners. of the studies described in Sections 2–4, responsiveness of a sub- Within this context, major questions involved discerning when and ject is altered if they have had previous experience with the UCS, how an individual can distinguish self from other, and whether if they have lived with or share kinship with their social partner, there was an ability to recognize that the perspective of another or if their social partner poses a concurrent threat. Moreover, in could be different from one’s own. At the highest level, cognitive humans, empathic responses are modulated by a subject’s per- approaches to empathy focused on language-based abstractions in ceived fairness of an individual suffering from a painful stimulus, which certain words could activate emotions in others because as well as by gender (Hein and Singer, 2008). In some situations they were relevant to a past experience. Moreover, some cogni- and species, a distinct form of cognitive empathy may be oper- tive approaches considered a role for compassion in the empathic ational, which requires an ability to distinguish that another can 1866 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 maintain a perspective distinct from one’s own. This ability to rec- such as fear. Importantly, this example of anisomorphic emotional ognize that the perspective of another individual is different from exchange is compatible with a perception–action mechanism, but one’s own is based upon a phenomenon known as ‘Theory of Mind’ particular emphasis is placed on the distinct affective experience (Farrow, 2007), a form of metacognition (see below). In this form, of each individual (viz., anger induces fear). Employing an affec- it is thought that cognitive empathy allows an individual to under- tive neuroscience approach, with its seven basic emotion systems, stand the emotional state of another without necessarily sharing allows one to identify and differentiate the affective experiences the same emotional state. In several instances, the expression of of animals (e.g., fear or happiness in one individual leads to fear or empathy is therefore determined not just by the ability to share happiness, respectively, in another individual). It is also important an affective experience, but also by factors that can modulate the to keep in mind that empathy can engender a string of subse- empathic experience. quent behavioral phenomena (e.g., escalated aggression, nurturing One way of broadly exemplifying the emotional versus cognitive or consolation behavior). aspects of empathy is through a comparison of two commonly used Observational learning studies serve as a foundation for study- clinical tests that evaluate psychosocial functioning in humans. ing empathy in rodents because they require an individual attend The ‘feigned distress test’ is used to assess whether an individual to the behavior of another individual. Numerous studies have attends to emotional changes in another. During the assessment, demonstrated that mice and rats are capable of learning from a clinician exclaims surprise in response to an accident, such as others (Zentall and Levine, 1972; Collins, 1988; Valsecchi et al., dropping papers on the floor or burning her fingers while lighting 2002; Carlier and Jamon, 2006). These studies typically involve a a candle on a cupcake. Normal children look towards the clinician’s task in which an observer experiences a demonstrator acquiring eyes and to the vicinity of the accident (i.e., papers on the floor, a foodstuff. In these experiments, hunger serves as the putative fingers), whereas some children on the autism spectrum express motivation for both the demonstrator performing the task and abnormal responses or fail to respond (Bacon et al., 1998; Dawson the observer attending to the behavior of the demonstrator. Thus, et al., 2004; Hutman et al., 2010). The feigned distress test thus iden- the demonstrator and the observer experience a common motiva- tifies whether an individual can detect and respond to an emotional tional state that does not necessarily involve a shared component. change in another. The term ‘shared’ in this review refers to an affective state in one Theory of Mind (ToM) is the ability of an individual to under- individual (‘demonstrator’) that can be communicated to another stand that another individual can have a separate experience from individual (‘observer’). Importantly, this type of shared experience one’s own. In one test of ToM (Wimmer and Perner, 1983), a human does not assume that a demonstrator intentionally communi- subject observes a scenario in which one doll (A) places a marble cates with an observer. Observational learning (including imitation in a basket and then leaves the room. Then, a second doll (B) enters and social facilitation) can be differentiated from empathy, which the room and removes the marble from the basket and places it occurs if a subject expresses a change in affect in response to an into an adjacent box. When doll A returns, the subject is asked to affective experience in a conspecific. By contrast, observational predict where doll A will search for the marble. Individuals with a learning occurs if changes in behavior represent modifications of healthy ToM identify the basket, understanding that doll A would task performance in the absence of emotional signals. Readers not have experienced doll B change the location of the marble. The are directed to excellent reviews by Heyes (1994) and Galef and individual can therefore infer, through cognitive re-representation, Giraldeau (2001) for additional perspectives on rodent observa- the perspective of another and it is thought that this ability is critical tional learning. for the expression of cognitive empathy. We focus on the affective component of empathy in rodents, we While empathy can include different levels of cognitive ability do not wish to imply that rodents are incapable of psychological that vary according to social, temporal and environmental contexts, processes related to cognitive empathy. Indeed, there is evidence emotions are a basic substrate for all empathic capacities. Indeed, that rodents possess the prerequisite skills for this ability, such a view of empathy that focuses on affective experience brings one as some forms of metacognition (Foote and Crystal, 2007). Rather, closer to the original translation of einfühlung or ‘empathy’ as a our focus on the affective aspects will build upon the solid frame- process of ‘feeling into’ another (Titchener, 1909). Within an exper- work of affective neuroscience and the basic emotional operating imental context, such as the rodent studies described below in systems of the mammalian brain (Panksepp, 1998). Specifically, a Sections 2–4, it may be useful to examine the features of a very prac- series of very recent studies is described that collectively demon- tical definition of empathy proposed by de Vignemont and Singer strate rodents are capable of (1) emotional contagion, (2) social (2006). In their definition, empathy occurs when an individual (A) modulation of associative learning, and (3) the shared affect com- experiences an affective state that is isomorphic to the affective ponent of empathy. We also describe where attempts have been state of another individual (B), and is elicited by observing or recall- made to identify the neural substrates that underlie these phenom- ing the respective expression of the affective state by individual ena. Taken together, these studies underscore that rodents possess B. Within this model, the affective state of individual B is there- a remarkable affective sensitivity to the emotional state of others, fore the source of emotional change in individual A (de Vignemont which in the future could be developed into robust experimen- and Singer, 2006). Some investigators would also allow empathy to tal models of psychosocial dysfunction relevant to disorders such include anisomorphic emotional responses between individuals A as autism (Bacon et al., 1998; Dawson et al., 2004; Hutman et al., and B (see below). However, the de Vignemont and Singer model 2010). provides a useful framework for studies of rodent affective states that are activated during a social exchange. 1.2. Foundational studies Emotional empathy requires that individuals share a feeling state based upon the same influence, even if these respective Prior to reviewing the more modern studies that are relevant experiences are temporally disconnected (i.e., recollection). This to rodent empathy, it is imperative to briefly discuss two stud- type of empathy can be conceptualized within the context of ies that were conducted over 45 years ago. Each of the studies is our understanding of the basic emotional systems of the mam- foundational in that they highlight the extent to which a rodent malian brain and how they generate affective feelings (Panksepp, can be attuned to the affective state of a social partner. In one 1998). For instance, an emotional experience of anger can engen- experiment (Church, 1959), the behavior of well-trained rats was der behavioral responsiveness (aggression) in a dominant male that assessed as they preformed a lever-pressing task and were con- subsequently induces an emotional experience in a subordinate, currently exposed to a conspecific that was being shocked. The J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1867 distress cues that were generated by a conspecific had a direct sup- algesia) compared to when pain was experienced individually. This pressive effect on the rate of lever pressing by subjects. The more effect was not specific to the type of noxious stimulus that was intriguing part of the experiment included an ‘emotional condition- experienced (acetic acid or formalin) or the resulting behavior ing’ component, during which subject rats and their social partners that was expressed (writhing or paw licking). (2) If experiencing simultaneously received a shock paired with a CS (i.e., panel lights different levels of pain together (e.g., each partner was simultane- in the testing apparatus). Control groups included subject rats that ously subjected to a different concentration of a noxious stimulus), experienced shocks that were presented at a different time than the behavior of each mouse was modulated by the level of pain the shocks delivered to their social partners. Another control group experienced by its social partner. For instance, if the forepaw of included rats were not conditioned. Following this phase of the one mouse was injected with a high concentration of formalin, it protocol, subjects were retested for lever pressing with intermit- exhibited less intense licking behavior if its partner had received tent co-exposure to conspecific distress and the CS. Subjects that a lower concentration of formalin. (3) Observing a conspecific in had previously been shocked contemporaneously with their social pain engendered the same amount of post-stimulus hyperalgesia partner during the conditioning phase of the experiment expressed as the direct experience of a painful stimulus. That is, the typi- a robust depression in lever pressing (relative to the other groups) cal increase in nociceptive sensitivity that mice undergo after a that persisted for 10 days. Subjects that were shocked at a different painful experience was also detected in individuals that had simply time than their social partner also decreased their lever pressing in observed mice in pain without experiencing the noxious stimulus response to co-presentation of the CS and conspecific distress, but themselves. their response was not as robust and it also returned to baseline Taken together, the Langford et al. (2006) findings demonstrated by the last day of testing. These results are particularly intriguing that mice are remarkably responsive to the level of pain expe- for two reasons: (1) Perception of social distress by a rat substan- rienced by other individuals. Control experiments that evaluated tially disrupted its behavior even though it was depleted of a vital different sensory modalities revealed that this form of social mod- resource (22-h food deprivation). (2) Shared experience of pain was ulation of pain was communicated by visual signals. Langford and a more effective modulator of behavior than was the consecutive co-workers argued that their behavioral findings were best con- experience of pain and perception of pain in others. Thus, the find- ceptualized within the context of emotional contagion, a process ings of Church (1959) were seminal because they demonstrated in which the behavioral state of an individual automatically acti- that a rat could identify whether its own emotional experience was vates the same state in nearby conspecifics. An analogous situation temporally coordinated with the same experience in a conspecific. to this contagion-type of responding also exists in rats, where In another study, Rice and Gainer (1962) asked whether a rat freezing by an observer is positively correlated with the amount could exhibit directed helping behavior to alleviate the distress of a of freezing expressed by a nearby demonstrator when tested in a conspecific. In the first part of the experiment, a subject was trained fear-conditioning paradigm (Wöhr and Schwarting, 2008). Notably, to press a lever to avoid delivery of a shock that was signaled by Langford et al. (2006) found that this process in mice was strongly a visual cue. Subsequently, subjects were tested with a distressed influenced by the degree of familiarity between social partners (also social partner that was suspended in the air via a hoist system (these see D’Amato and Pavone, 1993), a finding that resembles results distressed rats produced squeals in response to this procedure and from the human empathy literature. Moreover, the opposite effect their ‘wriggling’ behavior was visually perceptible to the subject). In (i.e., analgesia) was observed when the mouse in pain was paired this scenario, lever depression by the subject resulted in lowering of with an unfamiliar male that was not in pain (Langford et al., 2006). the conspecific and alleviation of its distress. Lever-pressing behav- This result was hypothesized to be due to the social threat that was ior of subjects was compared to a control group of rats that could posed by the unfamiliar male, a finding subsequently confirmed lower a Styrofoam block that was suspended in the air. Subjects that and shown to be dependent upon the hormonal status of the male had previous experience with conpecific distress expressed >10- partner (Langford et al., 2010). While the Langford et al. (2006) fold more responses to lower a distressed conspecific compared to study provided strong evidence that emotional contagion can be controls, whereas subjects that had never experienced conspecific studied in mice, a series of studies have demonstrated that com- distress expressed >3-fold more responses to lower a distressed munication between rodents modifies their future ability to learn conspecific relative to their respective control group. Collectively, about emotionally salient events. these findings demonstrated that rats would actively work (i.e., help) to reduce the distress of conspecifics, another phenomenon 3. Social interaction with a recently distressed conspecific that is relevant to empathy. modulates subsequent learning about environmental signals that predict distress 2. Social modulation of pain Four recent papers (two studies using mice and two using rats) Langford et al. (2006) conducted one of the first studies that revealed that a brief social exposure with a demonstrator modifies assessed empathy in rodents within the context of contemporary how an observer rodent subsequently performs in an associative neuroscience approaches. Their study adapted several traditional learning paradigm. Bredy and Barad (2009) demonstrated that the behavioral assays that are used to study pain in mice to include acquisition, retention and extinction of a cued-fear association a social component in which a ‘partner’ also experienced a nox- could be modified by a social interaction with a familiar conspecific ious stimulus. They asked if pain-related behaviors (e.g., writhing that was itself previously exposed to the same fear-conditioning after an intra-peritoneal injection of dilute acetic acid or paw lick- procedure. Specifically, an observer mouse experienced a 35-min ing after a subcutaneous injection of formalin) are sensitive to the social exposure in its home cage with an animal that had just behaviors of a social partner that also experiences pain. Individ- received three 2-s shocks (1 mA) that were each paired with a 2- ual mice were subjected to the pain-inducing manipulation, and min presentation of white noise (2-min inter-trial interval).Within their subsequent behavior was evaluated in real-time under exper- this behavioral context, the observer did not directly experience imental conditions of social isolation, a painful experience with a the demonstrator while it was distressed. Rather, the observer non-treated partner or a painful experience with a partner that was mouse interacted with the demonstrator in a familiar environment in pain at the same time. Their experimental results can be sum- at a time between fear conditioning of the demonstrator and its marized into three key findings: (1) If experiencing pain together, own conditioning. The observer mouse was then subjected to the mice expressed greater levels of pain-related behavior (i.e., hyper- same conditioning procedure as the demonstrator, and this expe- 1868 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 rience resulted in reduced acquisition of a fear response (freezing), Knapska and co-workers conducted additional in-depth analy- diminished ability to recall the CS–UCS association, and increased ses of social interactions with a distressed conspecific. When rats extinction of the fearful memory. All of these changes were indexed were reunited with a recently distressed conspecific, they directed by reduced freezing behavior in response to presentation of the >10-fold more allogrooming towards the individual compared to CS-only relative to observers that had interacted with a fear-naïve rats that were reunited with a non-distressed conspecific (Knapska mouse. It is important to note that the directionality of the observer et al., 2010). In rodents, allogrooming is a social activity, in which responses was somewhat unexpected, as it was predicted that a one individual repeatedly licks the head/neck/mouth/ear area of fearful demonstrator would respectively facilitate acquisition of a another. It is intriguing to speculate that the facilitation of this fear-induced memory and inhibit its extinction. behavior after detecting distress in another is an attempt to com- In a similar experiment, Knapska et al. (2010) found that the fort the afflicted individual, as has been previously suggested for ability of a rat to acquire and retain an active fear-conditioned primates (Preston and de Waal, 2002). Interestingly, Knapska et al. behavioral response (escape) in a two-way avoidance paradigm (2006) demonstrated a mutual and heightened induction of the was facilitated by previous interactions with a fearful demonstra- inducible transcription factor c-fos in the amygdala of distressed tor. In this behavioral paradigm, a demonstrator rat was exposed to demonstrator rats and observers following a brief social interac- ten 1-s shocks (1 mA) that were delivered across a 10-min period tion. Although heightened c-fos activity was detected in several (60 s inter-trial interval) and then the demonstrator was placed nuclei of the amygdala (e.g., lateral, basal, basal medial, medial and back into the home cage of the observer rat for a 10-min social central nuclei) of both distressed rats and the respective observers, interaction period. Following the social interaction, observer rats increased c-fos induction in the central nucleus of the amygdala were removed from the home cage and placed into a novel appa- differentiated these two groups. This finding suggests that partic- ratus where they were trained to avoid a foot shock by shuttling ular sectors of the amygdala are responsive to distress in others. back-and-forth between two chambers when cued by presenta- Overall, it appears that a coordinated set of changes in the amyg- tion of white noise. The results of this experiment demonstrated dala, a brain region implicated in numerous emotional processes, that when a rat was previously exposed to a distressed conspecific, occurs following the experience of self-distress and the experience the experience facilitated both the acquisition of avoidance behav- of distress in others. ior during the training phase of the experiment and an increased Using an approach similar to the two studies described above, retention of the fear memory, as indexed by increased avoidance Guzmán et al. (2009) found that observations of non-fearful behavior 24 h later. demonstrator mice modified subsequent learning about a CS–UCS Knapska et al. (2010) also demonstrated that interacting with association. In this study, observer mice were exposed to a novel a fear-conditioned demonstrator augmented recall of a passive context (3 min) on two successive days with a demonstrator behavioral response (freezing) in a contextual fear-conditioning that had previously received a 2-s shock (0.7 mA) in the same paradigm. In this paradigm, demonstrator and observer rats were context. The observers were then trained with the same condi- treated as described above, but observers were subsequently tioning paradigm. In this behavioral model, observations of fearful exposed to novel context for 4 min, and received a single 1-s foot demonstrators did not alter subsequent learning of the fear- shock (1 mA) at the transition between minutes 3 and 4 of the conditioning association by observers. However, if mice observed context exposure. Increased freezing behavior by the observer rats a non-fearful demonstrator, it inhibited their ability to recall the upon re-exposure to the context was detected 24 h later. The direc- CS–UCS association themselves. Previous experience with a fear- tionality of these results contrasts sharply with those of the Bredy naïve demonstrator therefore ‘buffered’ the likelihood that mice and Barad (2009) study, where experience with a fearful conspecific would become fear-conditioned. It is also worth noting that the reduced the expression of a fear memory. In developing a possible presence of a non-fearful partner during the presentation of CS explanation for this difference, it is important to consider how these that predicts application of a painful UCS retards the expression two experiments are similar. Careful control experiments by both of freezing in fear-conditioned rats (Kiyokawa et al., 2009). groups illustrated that changes in the ability of an observer to learn Unlike Bredy and Barad (2009) and Knapska et al. (2010), after social interaction with a distressed conspecific could not be observer mice in the Guzmán et al. (2009) study were able to derive explained by alterations in nociceptive sensitivity or generalized specific information about the experience of the respective demon- locomotion. For example, both studies demonstrated that interac- strators. While observers in all of the studies ultimately had to learn tion with a recently distressed animal did not alter the behavior of an association themselves, mice in the Guzmán et al. (2009) study observers in a hot-plate test of nociceptive sensitivity. Moreover, were sensitized to the appropriate context prior to the learning the differential ability of an observer to learn after social interac- experience. Thus, while the Bredy and Barad (2009) and Knapska tion with a distressed conspecific did not result from locomotor et al. (2010) studies illustrated that experience with social distress arousal. Bredy and Barad (2009) found that interaction with a dis- can subsequently modify the acquisition of a new behavior, the tressed mouse did not alter the exploratory response of observers Guzmán et al. (2009) study extended this finding further by demon- to a novel environment. This control experiment suggests that the strating that observers can make a direct connection between the increased freezing of observer mice was specific to the learning distress state of others and the specific environmental signals (i.e., experience. Knapska et al. (2010) found that social interaction with context) that are proximal to its occurrence. a distressed rat engendered faster acquisition of learned behavioral In a related experiment with rats, Bruchey et al. (2010) fear responses irrespective of whether the learned response required an conditioned demonstrators with 0.5-s shocks (0.7 mA) that were increase (escape) or decrease (freezing) in movement. It therefore each paired with a tone (three pairings), a protocol that engen- appears that an event during the social interaction with the dis- dered robust freezing responses to presentation of the CS 24 h tressed animal directly changed the propensity of the observers to later. By contrast, rats that were subjected to the same fear- learn a new association. Indeed, Bredy and Barad (2009) identified conditioning procedure, but with a less painful UCS (0.4 mA shock), ␤-phenylethylamine as a pheromone signal that communicated exhibited much less CS-induced freezing behavior. Importantly, distress to observer mice. Overall the findings of these two stud- if observer rats were allowed to socially interact with demon- ies demonstrate that the social experience of distress in others can strators that were simultaneously expressing conditioned freezing help rodents learn about fearful situations in the future. The expla- (what was referred to as ‘fear-conditioning by-proxy’), and then nation for the difference between mice and rats remains unknown subjected to the 0.4 mA-conditioning protocol, they expressed and warrants further experimental attention. freezing responses comparable to individuals that had underwent J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1869 the 0.7 mA-conditioning protocol. Thus, it appears that rats exhibit ultrasonic vocalizations (USVs). Naïve observers did not exhibit fear facilitated learning about cues that predict a painful stimulus if they responses to presentation of the CS or the conditioned responses have previously interacted with a conspecific that expresses fear to (freezing and 22 kHz-USVs) of the demonstrator. However, if the the same CS (see below). observer had previous experience with a UCS (three unsignaled Differences between the studies described in this section 1 mA-shocks), then paired testing resulted in robust observer freez- highlight several important issues that should be addressed sys- ing responses. The authors conducted a series of lesion/inactivation tematically in future studies of rodent empathy: To what extent experiments that targeted the medial geniculate nucleus (MGN) of is prior experience with the affect-inducing UCS required for the thalamus (i.e., auditory thalamus) and found that MGN activ- observers to respond? When do observers acquire information ity during the experience of unsignaled shocks was necessary for about the affective state of demonstrators (e.g., as the demonstra- future observer responses in the paired testing situation. Moreover, tor is becoming distressed by a UCS, after the UCS is removed, or as during paired testing the largest freezing responses were expressed the demonstrator changes its behavior in response to a CS that pre- by observer rats that had produced 22 kHz-USVs concurrently with dicts fear)? What factors play a role in directing changes in observer presentation of the unsignaled shocks. It was hypothesized that behavior towards a similar or different response relative to that of rats need to process their own 22 kHz-USVs during a painful expe- demonstrators? rience before they can respond appropriately to the distress state of demonstrators during paired testing. Thus, in this paradigm, a CS triggered a conditioned fear response in demonstrators, which 4. Social transfer of fear in turn engendered a fear response in ‘experienced’ observers. In another experiment, the Bruchey et al. (2010) study (also The experiments that have been reviewed so far illustrate that described in Section 2) found that rats acquire a freezing response rodents are immediately sensitive to the distress of others (Section simply by observing a demonstrator that is expressing a condi- 2), and that experience with conspecific distress can modulate how tioned fear response to a CS. In this experiment, demonstrator rats a rodent subsequently learns about environmental cues and con- were fear conditioned with the same protocol described in Sec- texts that predict fearful situations (Section 3). A critical question tion 2 (e.g., three tone-shock pairings). 24 h later, observers were regarding empathy in rodents is whether an individual can respond allowed to interact with demonstrators that were being presented to a CS that has been associated with the distress of a conspecific as with three playbacks of the CS-only. During a testing session 24 h if it were paired with the direct experience of a UCS. Below we later, observer rats expressed appreciable freezing responses when describe five studies that indicate rodents can be responsive in they were presented with the CS-only. Thus, in this experiment, a this way, which suggests that rodents are capable of experiencing CS that predicted distress in another engendered a fear response shared affect. as if the observer rat had experienced the UCS itself. Based on Employing a highly innovative and ethologically relevant the results of the Kim et al. (2010) study, it is possible that the paradigm, Kavaliers et al. (2003) asked whether a mouse could distress of demonstrators was communicated to observer rats via learn via social observation to avoid a predatory attack. Demonstra- 22 kHz-USVs. Interestingly, Bruchey et al. (2010) found that the tor mice were placed in a compartment lined with wood shavings responsiveness of observer rats to the CS was positively corre- and exposed to ≈50 biting flies for 30 min while an observer lated with the amount of social interaction they directed towards resided in an adjacent compartment that was separated by a mesh demonstrators during the observation session. The findings of Kim barrier. During this period, the demonstrator mouse exhibited et al. (2010) can be distinguished from those of Bruchey et al. several defensive behaviors in response to the biting flies and ulti- (2010), because observers did not require prior experience with mately buried itself under the bedding to avoid further attack. a UCS or presence of a demonstrator during testing (i.e., observers Testing included placing a demonstrator or observer in the same responded to the CS-only). context 24 h later and exposing them to flies that had their mouth- Employing a contextual conditioning paradigm, Jeon et al. parts removed, which eliminated the ability to bite. Demonstrators (2010) demonstrated that observer mice express freezing behavior expressed robust burying responses when they were re-exposed to when they are adjacent to a fearful demonstrator mouse receiv- flies, even though the UCS (i.e., biting) was not present. Importantly, ing repeated shocks over a 4-min period. This response by itself observer mice also avoided the non-biting flies by burying them- is consistent with emotional contagion because freezing of the selves, a response that was stimulus-specific because observers demonstrator and observer mice occurred at the same time. After did not attempt to avoid non-biting flies of a different species. the observer was placed back into the specific context where the Thus, the observer mice responded to the CS (i.e., flies) as if demonstrator had experienced the UCS 24 h earlier (equivalent to they anticipated a bite even though they had no direct experi- a long-term memory test), it also expressed freezing behavior, sug- ence with the UCS. Additional studies by Kavaliers et al. (2003) gesting that an association had been made between the distress revealed that exposure to non-biting flies following a single expe- of a conspecific and the CS. This subsequent effect was distinct rience with a distressed demonstrator engendered conditioned from emotional contagion because freezing behavior expressed by corticosterone release and analgesia in observers. Moreover, the the observer mouse occurred well after exposure to the distressed subsequent behavioral responses of an observer were dependent conspecific and observers had never experienced the UCS them- on visual cues and NMDA-signaling (Kavaliers et al., 2001) during selves. In other words, the freezing behavior of the observer mouse its experience with the demonstrator. was expressed as if it anticipated initiation of distress in another. Using a cue-conditioned fear paradigm, Kim et al. (2010) showed Freezing behavior in rodents is a direct, behavioral readout of fear that an observer rat can acquire a freezing response by observ- (LeDoux, 2000) and thus this finding reflects the social transfer of ing fear-conditioned demonstrators, but only if the observer has an emotional state from one mouse to another. had prior experience with the UCS itself. In this paradigm, demon- In earlier work (Chen et al., 2009), we developed a simi- strators were fear conditioned with ten 20-s tones that each lar experimental procedure in which observer mice experienced co-terminated with a 1-s shock (2 mA). During testing, demonstra- distressed conspecifics that underwent a series of tone-shock pair- tors were presented with the CS-only in the presence of an observer ings (Fig. 1). While the demonstrators were receiving the shock, rat (i.e., paired testing). The fear-conditioning procedure engen- observer mice did not express appreciable freezing behavior them- dered a robust fear response in demonstrators, evidenced by high selves, but they immediately oriented towards the demonstrator levels of CS-induced freezing and production of aversive 22 kHz- mice while the shock was being applied (Chen et al., 2009). This 1870 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875

Fig. 1. Conditioning apparatus to assess social transfer of fear among mice. Mice are habituated to the apparatus and then an observer (left) is exposed to a demonstrator (right) receiving 10 consecutive forward-pairings of a 30-s tone and a 2-s shock (90-s inter-trial interval). After 2 days of conditioning, the observer is placed on the side of the apparatus where the demonstrator was conditioned and it is then exposed to five presentations of the CS-only followed by five presentations of the CS–UCS contingency. The steel bars on the demonstrator side of the apparatus deliver electrical current whereas the bars on the observer side are inactive. Conditioning and testing are conducted during the dark phase of the circadian cycle under dim red illumination.

finding differs from what was reported for observer mice by Jeon ing the relationship between the strength of the UCS provided to et al. (2010), where individuals froze during the experience of the demonstrator and the expression of emotional contagion and distress in conspecifics. The discrepancy may be due to the fact shared affect. that Jeon et al. (2010) provided a very strong UCS to observers During the first half of the testing phase, 20 min after the last (20 2-s shocks [1 mA] delivered over a 4-min period) to demon- experience with demonstrators (equivalent to a short-term mem- strator mice, whereas our study utilized a more modest UCS (10 ory test), each observer mouse was individually placed back into 2-s shocks [0.5 mA] delivered over a 20-min period). Addition- the conditioning apparatus and their freezing behavior was exam- ally, observers and demonstrators were familiar in the Jeon et al. ined in response to the CS-only (the CS was not behaviorally salient (2010) study, whereas the mice in our study were unfamiliar (Chen prior to conditioning; Fig. 2a). Observer mice expressed freezing et al., 2009). While the reason for this difference between the behavior during playback of the CS as if they anticipated the ini- experiments is unknown, it does highlight important issues regard- tiation of distress in others (Fig. 2b). Thus, similar to Jeon et al.

Fig. 2. Freezing behavior of mice. (a) Without conditioning or observations of fearful demonstrators, mice from both the BALB and B6 genetic backgrounds expressed minimal freezing responses to the tone. Freezing was measured for 30 s during each CS presentation. (b) B6 mice expressed more freezing behavior than BALB mice to the CS if they had prior experience with fearful demonstrators. Freezing was measured for 30 s during each CS presentation on trials 1–5 and for 28 s on trial 6. (c) B6 mice that had previous experience with distressed demonstrators also expressed increased acquisition of freezing behavior relative to BALB mice when they were exposed to the CS–UCS contingency. Freezing was measured for 28 s during each CS presentation on trials 7–10. (d) BALB mice that did not have previous experience with distressed demonstrators expressed more freezing behavior than B6 mice to the CS when it was paired with a UCS. Freezing was measured for 28 s during each CS presentation on all trials. Asterisks represent a significant difference (P < 0.05) between the two genetic backgrounds. All data are presented as the mean ±standard error and were originally reported in Chen et al. (2009). J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1871

(2010), the observer mice in our study were responsive to a CS that between heart rate deceleration and empathic concern for others was associated with the induction of distress in others. Importantly, (Zahn-Waxler et al., 1995, also see Anastassiou-Hadjicharalambous this behavioral response were readily expressed by observer mice and Warden, 2008). from C57BL/6J (B6) strain, but not in individuals from the BALB/cJ Jeon et al. (2010) examined neuronal activity, specifically the (BALB) strain, which indicates that genetic background can influ- theta rhythm frequency (≈6 Hz) in the anterior cingulate cortex ence the degree to which a mouse is responsive to distress in others. (ACC) and the lateral amygdala (LA), and found that observing This strain-dependent difference is provocative because heritable conspecific distress synchronizes these oscillations. When these impairments in psychosocial functioning often include a compo- investigators used lidocaine to locally inactivate discrete brain nent where individuals fail to empathize (Goldenfeld et al., 2007; regions, they revealed that ACC activation was required to engender Hobson, 2007; Silani et al., 2008; Greimel et al., 2010). freezing to demonstrator distress, but not subsequent context- During the second half of the testing phase, observer mice were specific freezing behavior in observers. By contrast, LA activation themselves exposed to the CS–UCS contingency and their freezing was essential for both freezing responses during the experience behavior was evaluated. B6 mice also expressed increased acquisi- of conspecific distress and the subsequent expression of context- tion of the CS–UCS association relative to BALB observers (Fig. 2c). specific freezing behavior by observers. Thus, Jeon et al. (2010) By contrast, without the prior experience of distress in demon- proposed that the ACC was an essential brain region for the percep- strators, BALB mice expressed increased acquisition of the CS–UCS tion of conspecific distress in mice, whereas the LA was crucial for association relative to B6 mice (Fig. 2d). Thus, a genetic influence recalling this experience. As far as we know, the Jeon et al. (2010) on acquisition of a fearful memory (i.e., BALB > B6) was moderated study is the first of its kind in that it identified a neuronal sub- by prior experience with social distress (i.e., B6 > BALB). strate underlying a phenotype relevant to empathy in mice. Their We were able to reproduce the strain-dependent effect on freez- findings are especially important because the ACC, a brain region ing behavior in response to the CS-only and acquisition of the long implicated in the regulation of social processes (Panksepp, CS–UCS association by exposing observer mice to pairings of the 2003; Eisenberger and Lieberman, 2004), has been shown to acti- CS and playbacks of conspecific distress vocalizations (Chen et al., vate when humans detect distress in others (Singer et al., 2004; 2009). This finding suggests that vocal communication plays a cru- Jackson et al., 2005). cial role in the social transfer of fear among mice. A similar finding The experiments in this section are distinct from other studies has been demonstrated in rats, where the number of 22 kHz-USVs described in this review because they demonstrated that rodents emitted by a fearful demonstrator was positively associated with do not require direct experience with a CS–UCS contingency in the amount of fear that was expressed by an observer (Wöhr and order to express a behavior that reflects an affective response to Schwarting, 2008). Importantly, both BALB and B6 observers ori- presentation of the CS-only. Rather, observers gained an explicit ented to the distress vocalizations of demonstrator mice (Chen understanding of the CS–UCS association through a shared social et al., 2009), and age-matched mice from these strains exhibit experience. Many questions remain regarding the extent to which comparable thresholds for evoked auditory brainstem responses to this type of empathy is expressed in rodents. The studies that have stimuli that are presented at frequencies similar to the CS (1 kHz) been considered in this review implicate several key variables that and the distress vocalizations (≈20 kHz) (Zheng et al., 1999; Willott appear to modulate rodent sensitivity to distress in others; these et al., 1998). Furthermore, BALB mice were able to acquire the factors include familiarity with the demonstrator, prior experi- CS–UCS association when they were directly administered the UCS. ence with the UCS and the strength of the UCS delivered to the Thus, it appears that a difference in the hearing ability of mice demonstrator, among others. For the sake of clarity, we summarize from these two strains does not account for the difference in their the key points of similarity and distinction among the studies in sensitivity to conspecific distress. Table 1. In our view, it is likely that the expression of shared affect The BALB and B6 mouse strains have been developed as among rodents in any given behavioral paradigm is influenced by an experimental model of sociability by several laboratories a dynamic interaction between these factors. (Sankoorikal et al., 2006; Moy et al., 2007; Panksepp and Lahvis, 2007), and it has been suggested that the BALB mouse may be a valid mouse model of autism (Brodkin, 2007). By contrast, B6 mice 5. Synopsis and conclusions express high levels of gregariousness and social reward follow- ing social isolation, and are behaviorally sensitive to the induction All mammalian species are universally equipped with brains of distress in others (Panksepp and Lahvis, 2007; Panksepp et al., capable of generating felt, emotional experiences (Panksepp, 1998). 2007; Chen et al., 2009). In this regard, it is interesting to note that Although this assertion has been speculated for years (Darwin, a positive association between measures of empathic responsive- 1872/1998; Crowcroft, 1966), only recently has it become foun- ness and personality-trait indices of sociability is a common finding dational, largely through Jaak Panksepp’s seminal contributions. in the human empathy literature (Eisenberger et al., 2006). Thus, Within this context, it becomes almost axiomatic that rodents are further assessments of the inter-relationships between these social capable of empathy, a process where the affective feelings of one variables in B6 mice are warranted and may yield insights into the are conveyed to another and then generate the same feelings in that mechanisms underlying such phenotypic associations. individual. Consistent with the affective neuroscience approach, In addition to behavioral responsiveness, studies from both this argument is nevertheless a working hypothesis that can be fal- our lab (Chen et al., 2009) and the Shin lab (Jeon et al., 2010) sified. Perspectives on empathy in rodents should derive from an demonstrated that observer mice exhibit physiological correlates integration of behavioral, psychological, neural and evolutionary of empathy while they experience conspecific distress. For instance, approaches that are continuously updated based upon the available the heart rate of observer mice from the B6 strain increased imme- experimental evidence. diately at the beginning of the experience of conspecific distress Essentially all of the mouse studies described in this and then decreased below baseline with repeated exposures to review utilized individuals that were derived from the C57 distressed conspecifics. BALB mice also exhibited heart rate accel- line and it will be interesting to see if these findings gen- eration upon the induction of distress in conspecifics, but there was eralize to mice from additional strains. If empathy manifests no ensuing heart rate deceleration with repeated exposures to con- itself on a phenotypic continuum (see Decety et al., 2007 for specific distress. These results were particularly intriguing because a review), than we would anticipate that a graded empa- a series of studies in humans has demonstrated a relationship thy response should be found when comparing several strains. 1872 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 ] ↑↑ t during the , and the first Visual Sensory modality olfactory by Familiarity required? No Auditory Langford et al. (2006) Yes NotNo tested Not tested Enhanced pain experience required? No NoNoYes Visual Not tested Not tested Not testedNot tested (observers had 1 prior Olfactory experience w/shock) Yes Not tested Non- ↓ ↑ xcept for experiments 1 and 2 from freezing to cue Yes Not tested Not tested freezing to cue Nofreezing to cue Yes Not tested Not tested Not tested Auditory freezing to cue Yes Not tested Olfactory freezing to cuecue-based avoidance Yesfreezing to contextfreezing to Yesfreezing to context No No Olfactory Not tested avoidance to writhingpaw lickingthermal pain Yesfreezing to Yes cue (i.e., Yes Not testedorientation to freezing to cue that Not tested Visual freezing to context Yes Not tested Visual freezing to cue (i.e., ↑↑ ↑ ↑ ↓ ↓ ↑ behavior ↑ ↑ demonstrator distress, ↑ that predicts demonstrator distress ↑ non-biting flies of the same species ↑ ↓ ↑ behavior ↓ extinction) ↑ demonstrator distress, ↑ predicts demonstrator distress ↓ ↑ extinction) . 3 ] refers to an increase and decrease in behavioral responsiveness of the observer, respectively. [ × ↓ 3, 180-s Active 1-s shock 2-s shock ] and [ × ⇒ ⇒ ⇒ ↑ Social Social Kiyokawa et al. (2009) Demonstrator Extinction Extinction ⇒ ⇒ ⇒ ⇒ ⇒ Exposure to ⇒ 7, 30-180-s × 3, 60-s ITI 3, 120-s ITI 3, 120-s ITI × × × 2-s shock (0.7 mA) + context 0.5-s shock (0.4 mA) + cue ⇒ 3, 120-s ITI ⇒ × 3 Non-fearful demonstrator present during × ⇒ Demonstrator conditioned response to cue 1-s shock (1 mA) ITI response to cue (8 min continuous)testing during Demonstrator conditioned response to cue ITI testing 2-s shock (1 mA) interaction w/ demonstrator Social interaction w/ demonstrator avoidance training to 5-s shock (1 mA) + cue Social interaction w/ demonstrator (1 mA) + context Demonstrator distress + context Observer experiencenon-biting flies of the same species Observer response Observer sessions) 0.9% acetic acid injection + Demonstrator distress Demonstrator distress Social interaction w/ demonstrator (1 mA) training 2-s shock (1 mA) interaction w/ demonstrator Demonstrator distress + cue training ] refers to a transition between experiences. [ ⇒ 10, 20, 10, × × × 3, 3, 10 (2 10, 3, 120-s 3, 120-s 3, 120-s × × × × × × × , demonstrator pain experiences occurred prior to testing of the observer rodent. Demonstrators expressing a conditioned fear response were presen 10 extinction ⇒ 180-s ITI (2 mA) + cue 120-s ITI (0.7 mA) + cue 180-s ITI (0.7 mA) + cue (1 mA) + cue ITI (1 mA) + context 60-s ITI (1 mA) + context 10-s ITI (1 mA) + context 60-s ITI experience injection sessions), 90-s ITI (1 mA) + cue (1 mA) + cue injection ITI ITI trials (0.5 mA) + cue Jeon et al. (2010) and . Non-fearful demonstrators were present during the testing of observers in Rat 1-s shock Mouse 2-s shock Mouse Exposure to biting flies Demonstrator distress MouseRat Exposure to context No experience Non-fearful demonstrator in context (2 0.5-s shock (0.7 mA) Mouse 2-s shock (2) Mouse 2-s shock (1) Mouse(3) 2-s shock Mouse 2-s shock (2)(3) Mouse Mouse 1% formalin injection 0.9% acetic acid 5% formalin injection + Demonstrator distress (1) Mouse 0.9% acetic acid (1) Rat 1-s shock (2) Rat 1-s shock (2) Rat 0.5-s shock (1) Rat 0.5-s shock Kim et al. (2010) Chen et al. (2009) Kim et al. (2010) Bruchey et al. (2010) Bruchey et al. (2010) Bredy & Barad (2009) Knapska et al. (2010) Jeon et al. (2010) Knapska et al. (2010) Study (Experiment)Kavaliers et al. Species (2003) Demonstrator Langford et al. (2006) Langford et al. (2006) Guzmán et al. (2009) Kiyokawa et al. (2009) Langford et al. (2006) Bredy & Barad (2009) Bredy & Barad (2009) Chen et al. (2009) testing of observers in refers to an increase in the behavioral responsiveness of the observer relative to the first experiment in the study. Table 1 Summary of key variables from recent studies that focused on the detection and response of rodent observers to conspecific (demonstrator) distress. E [+] refers to a concurrent experience. [ITI] refers to inter-trial interval. [ observer responses in J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1873

Such a finding would be highly amenable to employing specific individual adjacent to the rat was a prior altruist, just that forward genetic approaches to understand mouse endophenotypes the subject had been a prior recipient of helping behavior from any that are relevant to empathy. On a related note, an association other individual. Rutte and Taborsky (2007) referred to this pro- between sociability and empathic responding (Chen et al., 2009; cess as ‘generalized reciprocity’ and suggested that positive affect Bruchey et al., 2010; Knapska et al., 2010) has not been thoroughly might be a mechanism underlying this phenomenon. In our view, studied in rodents and this too could be evaluated via assess- it is conceivable that positive social affect results from both the act ments of multiple strains. Another impending question entails how of helping and receiving, and thus shared pleasure might be a psy- the affective components of empathy in rodents can be modu- chological consequence of this type of cooperative behavior for rats lated by environmental and pharmacological factors. Several of (also see Schuster and Perelberg, 2004; de Waal et al., 2008). the studies described above demonstrated that greater familiarity Rough-and-tumble play is a robust and rewarding form of social between partners enhances empathic responding in mice (Langford activity that is expressed among young rats. In studies of rat play et al., 2006; Jeon et al., 2010), a finding that is consistent with behavior, pharmacological or environmental manipulation of the the human empathy literature (reviewed by Davis, 1994; Preston state of one rat in a dyad drastically alters the behavior of the other and de Waal, 2002). Moreover, given that many mental illnesses within this behavioral context (Poole and Fish, 1979; Humphreys include heightened or diminished abilities for empathy, includ- and Einon, 1981; Varlinskaya et al., 1999; Deak and Panksepp, ing schizophrenia (Derntl et al., 2009, Haker & Rossler, 2009), 2006), which suggests that playful rats are very attuned to the obsessive compulsive disorders (Fontenelle et al., 2009), anorexia- state of their partners. Indeed, adolescent rats prefer to be in the nervosa (Kucharska-Pietura et al., 2004), the neurochemical basis company of social partners that express greater amounts of 50 kHz- of empathic responding in rodents should be explored and candi- USVs (Panksepp and Burgdorf, 2003), which are a vocal reflection date molecules from a class of drugs termed ‘empathogens’ (Valea of positive social affect (Knutson et al., 2002). Thus, rats have a pref- and Hautefeuille, 2007) should be evaluated in rodents. erence for socially positive partners and appear to modulate their By definition, empathy requires that an individual be attuned play behavior accordingly when the behavior of their partner is dis- to the potential for an emotional experience in another. While cordant with their own. Taken together, these findings imply that the studies described in this review, as well as most work with rats enjoy play bouts with individuals that are able to share a similar humans, have focused on empathy within the context of negatively experience with them within the context of social interaction. valenced emotions (e.g., pain, sadness), the capacity for empathy A final and critical question regarding empathy in rodents is should also be considered when rodents experience positive emo- associated with its underlying neural circuitry. It will be interesting tions. The induction of negative emotional states has been easier to see how future brain studies involving rodent empathy corre- to study in animals because they have an immediate impact on the spond to the human counterpart experiments. Indeed, the initial likelihood for survival. However, ‘positive empathy’ happens every comparisons are promising (Jeon et al., 2010). However, rodent day. A good example of this in human life is when an individual brains can also be studied at levels of analysis that are either difficult answers a phone call and through the vocal prosody of a friend or unavailable in humans, apes and monkeys (e.g., brain region- or colleague, the individual immediately becomes invigorated by specific manipulation of neurochemical signaling, high-throughput excitement, even though he or she does not yet possess a propo- genetic approaches including altering the activity of single genes). It sitional knowledge of what underlies the excitement. The positive will be critical to integrate findings regarding the neural substrates aspects of empathy have been difficult to document in non-human of empathy in rodents with what is known about this process in animals, but there are some provocative recent examples that could humans. Consistent with the pioneering work of Jaak Panksepp be developed into rodent models of positive empathy. For exam- in affective neuroscience, as more sophisticated models of rodent ple, a recent study demonstrated that adolescent B6 mice become empathy emerge (including models of positive empathy), it will be ‘pseudosensitized’ to the locomotor-activating effect of morphine crucial to assess the extent to which the neural underpinnings of simply by observing other mice that had been treated with the drug empathy overlap with the basic emotional operating systems of the (Hodgson et al., 2010). Thus, mice were exposed to demonstra- mammalian brain (Panksepp, 1998). tor mice that were subjected to a typical morphine-sensitization The possibility of empathy in rodents beckons several addi- protocol across multiple days. On the test day, mice that had tional issues. First, if empathy exists in rodents, then it immediately received repeated treatments of morphine expressed the expected offers an opportunity to study this process at levels of analysis that hyperlocomotion phenotype in response to a standard dose of heretofore have been impossible. The field of mouse genetics is morphine. Remarkably, observer mice exhibited an enhanced loco- continually evolving, but with the technologies that are currently motor response to the same morphine dose as if they had previous available, it is realistic that the genetic substrates underlying the experience with the drug. The doses of morphine used by Hodgson ability to detect and respond to distress in others can be discovered et al. (2010) were within that range that is typically used to measure and eventually developed into targets for pharmacotherapeutic reward in conditioned place preference studies. Since locomotor intervention. At a still greater level, Panksepp’s approach to affec- sensitization is thought to reflect a state of reward, it is intriguing tive neuroscience has provoked us to consider that several basic to speculate that the observer mice shared the reward state with affective states can be experienced by all mammalian species. If demonstrator mice during the sensitization protocol, and that the some of these emotional experiences can be induced in animals via enhanced locomotor response of observers during testing was a other individuals (i.e., shared affect) then many of us will be forced reflection of this shared experience. It is important to note that to reconsider our own worldview of animal nature. Similarly, the the actions of a pharmacological compound, such as morphine, in recognition that rodents possess a capacity for empathy will require the brain may be quite different than a more natural stimulus, and scientists to carefully evaluate their own research practices in an the mechanisms underlying pseudosensitization may thus be very attempt to reduce the potential for excessive communication of different than those described in Sections 2–4. pain or fear between individuals. Another case where positive empathy might come to play is in an experimental paradigm where rats are trained to cooperate to pull in a baited platform to their cage to procure an oat flake References (Rutte and Taborsky, 2007). A subject would pull for an unfamiliar Anastassiou-Hadjicharalambous, X., Warden, D., 2008. Children’s heart rate and conspecific that was adjacent to its cage as long as the act had been vicariously aroused affect in response to others’ differing emotional experiences. previously performed for the subject; it did not matter whether the J. Open Psychol. 1, 78–83. 1874 J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875

Bacon, A.L., Fein, D., Morris, R., Waterhouse, L., Allen, D., 1998. The responses of Hoffman, M.L., 1981. Is altruism part of human nature? J. Pers. Soc. Psychol. 40, autistic children to the distress of others. J. Autism Dev. Disord. 28, 129–142. 121–137. Bredy, T.W., Barad, M., 2009. Social modulation of associative fear learning by Hoffman, M.L., 1987. The contribution of empathy to justice and moral judgement. pheromone communication. Learn. Mem. 16, 12–18. In: Eisenberg, N., Strayer, J. (Eds.), Empathy and its Development. Cambridge Brodkin, E.S., 2007. BALB/c mice: low sociability and other phenotypes that may be University Press, United Kingdom, pp. 47–80. relevant to autism. Behav. Brain Res. 176, 53–65. Humphreys, A.P., Einon, D.F., 1981. Play as a reinforcer for maze-learning in juvenile Bruchey, A.K., Jones, C.E., Monfils, M.H., 2010. Fear conditioning by-proxy: social rats. Anim. Behav. 29, 259–270. transmission of fear during memory retreival. Behav. Brain Res. 214, 80–84. Hutman, T., Rozga, A., DeLaurentis, A.D., Barnwell, J.M., Sugar, C.A., Sigman, M., 2010. Carlier, P., Jamon, M., 2006. Observational learning in C57BL/6j mice. Behav. Brain Response to distress in infants at risk for autism: a prospective longitudinal Res. 174, 125–131. study. J. Child Psychol. Psychiatry 51, 1010–1020. Chen, Q., Panksepp, J.B., Lahvis, G.P., 2009. Empathy is moderated by genetic back- Jackson, P.L., Meltzoff, A.N., Decety, J., 2005. How do we perceive the pain of oth- ground in mice. PLoS One 4, e4387. ers? A window into the neural processes involved in empathy. NeuroImage 24, Church, R.M., 1959. Emotional reactions of rats to the pain of others. J. Comp. Physiol. 771–779. Psychol. 52, 132–134. Jeon, D., Kim, S., Chetana, M., Jo, D., Ruley, H.E., Lin, S.Y., Rabah, D., Kinet, J.P., Shin, Collins, R.L., 1988. Observational learning of a left-right behavioral asymmetry in H.S., 2010. Observational fear learning involves affective pain system and Cav1.2 mice (Mus musculus). J. Comp. Psychol. 102, 222–224. Ca2+ channels in ACC. Nat. Neurosci. 13, 482–488. Crowcroft, P., 1966. Mice All Over. G.T. Foulis and Co. Ltd., United Kingdom. Kavaliers, M., Colwell, D.D., Choleris, E., 2001. NMDA mediated social learning of Darwin, C., 1872/1998. The Expression of the Emotions in Man and Animals. Oxford fear-induced conditioned analgesia to biting flies. Neuroreport 12, 663–667. University Press, Oxford, 473 pp. Kavaliers, M., Colwell, D.D., Choleris, E., 2003. Learning to fear and cope with a nat- Davis, M.H., 1994. Empathy. A Social Psychological Approach. Brown and Benchmark ural stressor: individually and socially acquired corticosterone and avoidance Publishers, Madison, 260 pp. responses to biting flies. Horm. Behav. 43, 99–107. Dawson, G., Toth, K., Abbott, R., Osterling, J., Munson, J., Estes, A., Liaw, J., 2004. Early Kim, E.J., Kim, E.S., Covey, E., Kim, J.J., 2010. Social transmission of fear in rats: the social attention impairments in autism: social orienting, joint attention, and role of 22-kHz ultrasonic distress vocalization. PLoS One 5, e15077. attention to distress. Dev. Psychol. 40, 271–283. Kiyokawa, Y., Takeuchi, Y., Nishihara, M., Mori, Y., 2009. Main olfactory system medi- D’Amato, F.R., Pavone, F., 1993. Endogenous opioids: a proximate reward mechanism ates social buffering of conditioned fear responses in male rats. Euro. J. Neurosci. for kin selection? Behav. Neural Biol. 60, 79–83. 29, 777–785. Deak, T., Panksepp, J., 2006. Play behavior in rats pretreated with scopolamine: Knapska, E., Nikolaev, E., Boguszewski, P., Walasek, G., Blaszczyk, J., et al., 2006. increased play solicitation by the non-injected partner. Physiol. Behav. 87, Between-subject transfer of emotional information evokes specific pattern of 120–125. amygdala activation. Proc. Natl. Acad. Sci. U.S.A. 103, 3858–3862. Decety, J., 2010. The neurodevelopment of empathy in humans. Dev. Neurosci. 32, Knapska, E., Mikosz, M., Werka, T., Maren, S., 2010. Social modulation of learning in 257–267. rats. Learn. Mem. 17, 35–42. Decety, J., Jackson, P.L., 2004. The functional architecture of human empathy. Behav. Knutson, B., Burgdorf, J., Panksepp, J., 2002. Ultrasonic vocalizations as indices of Cogn. Neurosci. Rev. 3, 71–100. affective states in rats. Psychol. Bull. 128, 961–977. Decety, J., Jackson, P.L., Brunet, E., 2007. The cognitive neuropsychology of empathy. Kucharska-Pietura, K., Nikolaou, V., Masiak, M., Treasure, J., 2004. The recognition of In: Farrow, T.F.D., Woodruff, P.W.R. (Eds.), Empathy in Mental Illness. Cambridge emotion in the faces and voice of anorexia nervosa. Intl. J. Eating Disorders 35, University Press, United Kingdom, pp. 239–260. 42–47. Derntl, B., Finkelmeyer, A., Toygar, T.K., Hulsmann, A., Schneider, F., Falkenberg, Langford, D.J., Crager, S.E., Shehzad, Z., Smith, S.B., Sotocinal, S.G., et al., 2006. Social D.I., Habel, U., 2009. Generalized deficit in all core components of empathy in modulation of pain as evidence for empathy in mice. Science 312, 1967–1970. schizophrenia. Schizo. Res. 108, 197–206. Langford, D.J., Tuttle, A.H., Briscoe, C., Harvey-Lewis, C., Baran, I., et al., 2010. Vary- de Vignemont, F., Singer, T., 2006. The empathic bra: how, when and why? Trends ing perceived social threat modulates pain behavior in male mice. J. Pain 12, Cogn. Sci. 10, 435–441. 125–132. de Waal, F.B., 2008. Putting the altruism back into altruism: the evolution of empa- LeDoux, J.E., 2000. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184. thy. Annu. Rev. Psychol. 59, 279–300. Lipps, T., 1903. Einfühlung, innere Nachahmung, und Organepfindungen. Archiv für de Waal, F.B., Leimgruber, K., Greenberg, A.R., 2008. Giving is self-rewarding for die gesamte Psychologie 1, 185–204. monkeys. Proc. Natl. Acad. Sci. U.S.A. 105, 13685–13689. MacLean, P.D., 1967. The brain in relation to empathy and medical education. J. Nerv. Eisenberger, N.I., Lieberman, M.D., 2004. Why rejection hurts: a common neural Mental Dis. 144, 374–382. alarm system for physical and social pain. Trends Cogn. Sci. 8, 294–300. Moy, S.S., Nadler, J.J., Young, N.B., Perez, A., Holloway, L.P., et al., 2007. Mouse behav- Eisenberger, N., Fabes, R.A., Spinrad, T.L., 2006. Prosocial development. In: Eisenberg, ioral tasks relevant to autism: phenotypes of 10 inbred strains. Behav. Brain Res. N. (Ed.), Handbook of Child Psychology, 3. John Wiley & Sons Inc., New Jersey, 176, 4–20. pp. 646–718. Panksepp, J., 1998. Affective Neuroscience: The Foundation of Human and Animal Farrow, T.F.D., 2007. Neuroimaging of empathy. In: Farrow, T.F.D., Woodruff, P.W.R. Emotions. Oxford University Press, New York, 480 pp. (Eds.), Empathy in Mental Illness. Cambridge University Press, United Kingdom, Panksepp, J., 2003. Feeling the pain of social loss. Science 302, 237–239. pp. 199–216. Panksepp, J., 2005. Affective consciousness: core emotional feelings in animals and Farrow, T.F.D., Woodruff, P.W.R., 2007. Empathy in Mental Illness. Cambridge Uni- humans. Conscious. Cogn. 14, 30–80. versity Press, United Kingdom. Panksepp, J., Burgdorf, J., 2003. Laughing rats and the evolutionary antecedents of Fontenelle, L.F., Soares, I.D., Miele, F., Borges, M.C., Prazeres, A.M., Range, B.P., human joy? Physiol. Behav. 79, 533–547. Moll, J., 2009. Empathy and symptoms dimensions of patients with obsessive- Panksepp, J.B., Lahvis, G.P., 2007. Social reward among juvenile mice. Genes Brain compulsive disorder. J. Psychiat. Res. 43, 455–463. Behav. 6, 661–671. Foote, A.L., Crystal, J.D., 2007. Metacognition in the rat. Curr. Biol. 17, 551–555. Panksepp, J.B., Jochman, K.A., Kim, J.U., Koy, J.J., Wilson, E.D., et al., 2007. Affiliative Galef Jr., B.G., Giraldeau, L.A., 2001. Social influences on foraging in vertebrates: behavior, ultrasonic communication and social reward are influenced by genetic causal mechanisms and adaptive functions. Anim. Behav. 61, 3–15. variation in adolescent mice. PLoS One 2, e351. Goldenfeld, N., Baron-Cohen, S., Wheelwright, S., Ashwin, C., Chakrabarti, B., 2007. Poole, T.B., Fish, J., 1979. An investigation of individual, age and sexual differences Empathizing and systematizing in males, females and autism: a test of the neu- in the play of Rattus norvegicus (Mammalia:Rodentia). J. Zool. 179, 249–259. ral competition theory. In: Farrow, T.F.D., Woodruff, P.W.R. (Eds.), Empathy in Preston, S.D., de Waal, F.B.M., 2002. Empathy: its ultimate and proximate bases. Mental Illness. Cambridge University Press, United Kingdom, pp. 322–334. Behav. Brain Sci. 25, 1–72. Greimel, E., Schulte-Ruther, M., Kircher, T., Kamp-Becker, I., Remschmidt, H., Fink, Rice, G.E., Gainer, P., 1962. Altruism in the albino rat. J. Comp. Physiol. Psychol. 55, G.R., Herpertz-Dahlmann, B., Konrad, K., 2010. Neural mechanisms of empathy 123–125. in adolescents with autism spectrum disorder and their fathers. Neuroimage 49, Rutte, C., Taborsky, M., 2007. Generalized reciprocity in rats. PLoS Biol. 5, e196. 1055–1065. Sankoorikal, G.M., Kaercher, K.A., Boon, C.J., Lee, J.K., Brodkin, E.S., 2006. A mouse Guzmán, Y.F., Tronson, N.C., Guedea, A., Huh, K.H., Gao, C., Radulovic, J., 2009. Social model system for genetic analysis of sociability: C57BL/6J versus BALB/cJ inbred modeling of conditioned fear in mice by non-fearful conspecifics. Behav. Brain. mouse strains. Biol. Psychiat. 59, 415–423. Res. 201, 73–78. Schuster, R., Perelberg, A., 2004. Why cooperate? An economic perspective is not Haker, H., Rossler, W., 2009. Empathy in schizophrenia: impaired resonance. Euro. enough. Behav. Processes 66, 261–277. Arch. Psychiat. Clinical Neurosci. 259, 352–361. Silani, G., Bird, G., Brindley, R., Singer, T., Frith, C., Frith, U., 2008. Lev- Hein, G., Singer, T., 2008. I feel how you feel but not always: the empathic brain and els of emotional awareness and autism: an fMRI study. Soc. Neurosci. 3, its modulation. Curr. Opin. Neurobiol. 18, 153–158. 97–112. Heyes, C.M., 1994. Social learning in animals: categories and mechanisms. Biol. Rev. Silk, J.B., 2007. Empathy, sympathy and prosocial preferences in primates. In: Dun- Camb. Philos. Soc 69, 207–231. bar, R.I.M., Barrett, L. (Eds.), Oxford Handbook of Evolutionary Psychology. Hobson, P., 2007. Empathy and autism. In: Farrow, T.F.D., Woodruff, P.W.R. (Eds.), Oxford University Press, United Kingdom, pp. 115–126. Empathy in Mental Illness. Cambridge University Press, United Kingdom, pp. Singer, T., Lamm, C., 2009. The social neuroscience of empathy. Ann. N. Y. Acad. Sci. 126–141. 1156, 81–96. Hodgson, S.R., Hofford, R.S., Roberts, K.W., Wellman, P.J., Eitan, S., 2010. Socially Singer, T., Seymour, B., O’Doherty, J., Kaube, H., Dolan, R.J., Frith, C.D., 2004. Empathy induced morphine pseudosensitization in adolescent mice. Behav. Pharmacol. for pain involves the affective but not sensory components of pain. Science 303, 21, 112–120. 1157–1162. Hoffman, M.L., 1975. Developmental synthesis of affect and cognition and its inter- Titchener, E., 1909. Elemental Psychology of the Thought Processes. Macmillan, New play for altruistic motivation. Dev. Psychol. 11, 607–622. York. J.B. Panksepp, G.P. Lahvis / Neuroscience and Biobehavioral Reviews 35 (2011) 1864–1875 1875

Valea, D., Hautefeuille, M., 2007. Empathogenic agents: their use, abuse, mechanism Wimmer, H., Perner, J., 1983. Beliefs about beliefs: representation and constrain- of action and addiction potential. In: Farrow, T.F.D., Woodruff, P.W.R. (Eds.), ing function of wrong beliefs in young children’s understanding of deception. Empathy In Mental Illness. Cambridge University Press, United Kingdom, pp. Cognition 13, 103–128. 199–216. Wöhr, M., Schwarting, R.K.W., 2008. Ultrasonic calling during fear conditioning in Valsecchi, P., Bosellini, I., Sabatini, F., Mainardi, M., Fiorito, G., 2002. Behavioral anal- the rat: no evidence for an audience effect. Anim. Behav. 76, 749–760. ysis of social effects on the problem-solving ability in the house mouse. Ethology Zahn-Waxler, C., Cole, P.M., Welsh, J.D., Fox, N.A., 1995. Psychophysiological corre- 108, 1115–1132. lates of empathy and prosocial behaviors in preschool children with behavior Varlinskaya, E.I., Spear, L.P., Spear, N.E., 1999. Social behavior and social motivation problems. Dev. Psychopathol. 7, 27–48. in adolescent rats: role of housing conditions and partner’s activity. Physiol. Zentall, T.R., Levine, J.M., 1972. Observational learning and social facilitation in the Behav. 67, 475–482. rat. Science 178, 1220–1221. Willott, J.F., Turner, J.G., Carlson, S., Ding, D., Seegers Bross, L., Falls, W.A., 1998. The Zheng, Q.Y., Johnson, K.R., Erway, L.C., 1999. Assessment of hearing in 80 strains of BALB/c mouse as an animal model for progressive sensorineural hearing loss. mice by ABR threshold analyses. Hear. Res. 130, 94–107. Hear Res. 115, 162–174.